Energy-aware task scheduling strategies with QoS constraint for green computing in cloud data centers

Energy optimization with Quality-of-Service (QoS) constraint has become a timely and significant challenge for the cloud datacenters. In this paper, a hardware and software collaborative optimization strategy is implemented to minimize the energy cost while satisfying the time constraint of the cloud-computing datacenters. In the hardware aspect, a DVFS-capable CPU/GPU/FPGA heterogeneous cloud infrastructure is built. This infrastructure has high flexibility, and can adjust its hardware characteristics dynamically in terms of the software run-time contexts, so that a hardware platform which matches the software can be built. Based on this hardware platform, the cloud applications can be executed more efficiently with less energy cost. In the software aspect, the deadline-aware energy-efficient task scheduling algorithms are investigated. Different from the traditional approaches which search for the optimal scheduling solution by the heuristic approaches, a new scheduling approach based on the improved Mathematical Morphology (MM) algorithm is investigated in this paper. To evaluate the performance of our work, we calculated the energy cost of the Fourier transform (FT) and Gaussian elimination (GE) applications on the homogeneous and heterogeneous cloud computing platforms by applying the GA and MM algorithms, respectively. The results proved the MM algorithms running on the DVFS-capable heterogeneous cloud infrastructure could decrease the energy cost of the FT application and GE application respectively by 24.7% and 37.8%, if compared with the GA algorithm running on the DVFS-incapable homogeneous cloud infrastructure.